Hi guys,

well – yesterday was one of those days you whish you simply stayed in bed

Early in the morning and before work starts, I recognized that the newer WD Green 1TB HDD I’ve just added together with a Seagate 1TB HDD to the server’s RAID (s. last post) decided to drop out of the RAID. Urgs! Not good for a 7:30am session. Trying to get this drive back into the RAID failed. Seems that it’s really broken. I’m somewhat happy that it’s a hardware based RAID5, so I could do a backup of the data from the server.

Bad enough.

Around noon, I suddenly heard an “uncool” sound from one of the two RAID0 disks inside my Workstation. I’d just installed it the same morning, switching completely from OSX to Windows 7 on my Workstation. Right after hearing this sound (a “huiiiiii” “clack”), a bluescreen appears on the screen. Uh oh! Rebooting the system showed me, that the HDD was dropped out of the RAID0 and so – of course – all the work from the day was gone. Luckily I just had installed things, because I have to work and have no time to play with the Workstation the same time.

The 2nd broken harddisk was a Samsung 250GB SATA-II HDD with 7200upm and 16MB of cache. I checked the date it was produced: May 2008. Oh!? I found the invoice instantly (thanks to my wifes very well sorted folders ). I still had warranty.

So I went to the computer store, where I bought it on 08.08.08. No that’s no joke! It was really bought that day . What I like at this store is, that they don’t do those buerocratical dramas like smaller shops. For example: I brought my broken ArcticCooling PSU back to the other dealer and they send it to the manufacturer – so I will have to wait 2 weeks to see if I get money back or not. Anyway. So they simply exchange the HDD if I wanted to, but I had something else in mind. Getting a credit instead of getting the same damn disk again. Don’t understand me wrong. I like and nearly 100% used Samsung HDDs in the past, but today, I’m much more happy with WDs. So I decided to get 3 of their WD Caviar Blue, 640GB, 16MB cache and 7200upm instead of the same 250GB Samsung disk. Funny enough, I got back 45 Euros for the broken drive. The new price last year was 49 Euro. Great deal

When I came back home, I’ve put all 3 new WD Caviar Blues into the Workstation and created an onboard RAID5 (ICH10R) and I must say, I’m really happy with it. The speed is of course not as good as if I had used my Areca RAID controller, but it’s a bit faster than a single drive and it’s reliabel. And that’s what counts for me.

Wondering why I didn’t got 3 or 4 larger ones? Well, that’s easily explained: I don’t need that much storage in the Workstation. I’m happy with the 1.2TB I now have after substracting the 3rd HDD for RAID5 parity. 4 were simply too expensive in the moment, but I still have 3 more SATA slots available and 3 more slots for additional HDDs

Now, that the RAID in the server was broken, I had to recreate the whole RAID and I thought it’s maybe a better idea to also remove the Seagate HDD from the RAIDset, too. The external 1TB backup drive is filled by 99% and I need a 2nd one, so I simply put this Seagate 1TB HDD into an external case and have more space for the backup.

While being more or less happy with the Debian as server, but still pretty annoyed with user and system rights and just got the Windows Server 2008 R2 x64 via MSDN, I wanted to try this of course, as it’s based on Windows 7 and no more on Vista and should be WAY faster.

The first issue was: How the hell to get the server OS to recognize the Areca controller?! Everytime I load the drivers, I get a message that it’s not the right one.

Hmm… I tried it with the German and the English versions (Full). I also tried it with USB and floppy disk (yeah those fancy 3.5″ 1.44MB plastic things I still have a USB floppy for those cases and it’s cool as it is sooo retro . But – unfortunately – no success. I was in contact with the Areca support for some days and I had the feeling that they were just using Evaluation copys of the Windows Server 2008 R2 in their labs. So I went to the Microsoft website and downloaded the Evaluation of the English version. Voilá – it works! And I was able to use my serial numbers with this as well. Ok – so I don’t want to use the English version and got the German Evaluation version, a bit unsure if the driver and my official serial number will work here. It does!

Very strange, but that’s the currently only work around. I think I will get a newer reply from Areca in the next days telling me to do this workaround

By the way: It’s a shame that I’ve paid 400 Euro for a controller and they still have so much issues with Windows Server 2008 R2 and ALSO with Windows 7. Luckily I’m using no Areca controller in the Workstation. Maybe the next one will be an DELL/LSI Perc/i6/7 or an Adaptec one.

So that’s it for today Enjoy the weekend!

Ciao
Dennis

Tweet This Post Digg This Facebook Reddit

Hey guys,

having a reliable storage is nice. Having it in well designed case is even better. Until last week I had stored my main server “Belldandy” or “Hadron” which know listens to the name “Spike” (request of my wife ) in a 4U 19″ case. This was nice as it fits into the 19″ rack in the basement, but it was horrible if you need to change something. And I just speak about adding a new card (like ISDN or network card). Because I had no rails installed, this makes the whole thing a bit more complicated. I had to remove the whole system from the rack everytime which is kinda annoying. On top of this, the new bought (1 1/2 years ago) cases were never really good. Yes – you can add up to 6 HDDs internally, but you need to fix every HDD with at least 6 screws plus the 5.25″ cage was accidently broken because of bad quality.

My wish was to use the Areca 1220 controller completely and not “just” 6 of the 8 ports. So I’ve got 2 more 1TB HDDs. A few weeks ago I’d bought a new Chieftech MESH case initially to put the server into the home office room, but it turns out, that it’s simply too hot here when I run this system side by side with the Workstation, the ThinkPad T61 for work and my MacBook the same time. So I decided to use the 4U 19″ case. Bad mistake. Since this case just had 6 internal 3.5″ bays and the Chieftec has 8 of them, the decision was clear – the server has to be moved back to the huge case.

Another thing that was (for my surprise) broken way too early was the ArcticCooling 80+ power supply that I’d used in the server. The switch does not work. I’m honestly not sure if it ever worked, I had added the power cord to the supply and turned on the machine then in the past. I hope I can give it back. I will try that today. As a replacement, I’ve ordered a luxury power supply. For a machine that runs 24/7 mostly, I want to have something reliable and also highly energy efficient. My choice: the Corsair HX750W with cable management, 90% Energy efficiency and a BUNCH of connectors. 3×4 SATA cables came with the supply and 2×4 MOLEX ones as well as 4x PCIexpress. Not that I need all of them. I’d stick with 2×4 SATA & 1×4 MOLEX to run the server. The graphics card is an pretty old but reliable ATI PCI card which is fine for a server.

I’ve mounted it all to the newer case and the result looks like this. I’m still a bit in love with the fancy stack of HDDs

Here are some “screenshots” of the Areca controllers setup when setting up the storage. The initialization (as background process) for the 6880GB Volume took around about 25-26 hours! Luckily you can fully use it at the same time and install your system meanwhile.

Last, but not least, here are the details of the new server:

- Intel Core 2 Duo, 2.4GHz (E6600)
- 8GB DDR2-800 Kingston RAM
- 8TB RAID5 Storage (8x 1TB HDDs, 7x WD Green, 1x Seagate)
- 120GB System Volume (RAID5)
- 6.8TB Storage Volume (RAID5)
- Areca 1220 PCIe x8 RAID controller
- Onboard GBit network
- HFC-S ISDN card PCI
- AVM ISDN card PCI
- ATI PCI graphic card
- Corsair HX750W 90% EE power supply
- OS: GNU/Debian Linux “Lenny” 5.0.3

If you have any questions, feel free to post a comment or send me an eMail

Ciao
Dennis

Tweet This Post Digg This Facebook Reddit

Hi guys,

a few days ago, I received an eMail from StarWind Software who are asking me if I were interested in testing their product – the Enterprise Server. So they’ve send me a licence to play around with it and I took the time to do it.

By the way: It seems like they are on a war path at the moment, they also asked Sally, who is the fiance of Stefan to test it, too. Read there review on the linkes pages

Overview about the systems I use for the test:

Server (Target):
- Windows Server 2008 Standard Edition 64bit SP2
- Intel Core 2 Duo E6600, 2.4GHz (2x)
- 8GB DDR2-800 RAM
- 6x 1TB WD Green HDDs, managed as RAID5 by an Areca 1220 PCIe x8 hardware RAID controller

Client (Initiator) 1:
- Mac OS X 10.6.1 Snow Leopard
- Intel Core 2 Duo, 2GHz (2x)
- 4GB DDR3-1066 RAM
- 160GB SATA HDD

Client (Initiator) 2:
- Windows 7 Ultimate x64
- Intel Core 2 Quad Q9650, 3GHz (4x)
- 8GB DDR2-800 RAM
- 2x 200GB Samsung HDDs in an “onboard” RAID0

Description:

iSCSI – that sounds like a lot of fun, right? Well, it is! iSCSI is a kind of a network protocol that includes the SCSI protocol and allows to use it inside a network. Huh – complicated stuff, I know.

Well, to explain it easily: You can create a “real” harddisk inside your computer by using your regular OS tool like the Harddisk-Utility in Mac OS X or the Device-Manager (diskmgmr.msc) in any Windows system. The client system itself needs to be ready to do this.

Mac OS X is itself NOT able to do the initator job, as far as I know, but there’s a free System Preferences Plugin with the name globalSAN iSCSI Initiator for the Mac and they give it away for free (free like free beer).

Windows 7 (and also Windows Vista previously) are able to use this server-based-service from scratch and install the iSCSI initiator with the system installation.

Installation:

My very first step (of course) was the installation of the Enterprise Server software on my Windows Server 2008. The installation was quiet easy and it turns out, that the software itself is also quiet easy to use.

The StarWind Enterprise Server console on the Windows Server

Creating an image file is just one of a lot various options

The next step was to create an image file as base for the test-HFS+ harddisk I wanted to create.

Snow Leopard connects to the iSCSI server:

That was “easy peasy” and a few minutes later, I had connected the globalSAN iSCSI initiator with the server. (Note: the globalSAN software was installed earlier).

The globalSAN iSCSI initiator for Mac is seeing the iSCSI device. Ready to rumble

The Disk Utility now also recognize the iSCSI device and allows me to do whatever I would do to a regular harddisk. In this case, I’ve removed my Test HFS+ partition and created a new one which I’ve called “Klein2″ for this test.

CMD + I shows this volume information (sorry, in German on this screenshot)

Windows 7 connects to the iSCSI server:

For my surprise it was a bit easier to use Windows 7 than to use Mac OS X – because I had just to put in the servers hostname “belldandy” and the targets where found on the fly.

The Windows 7 iSCSI initiator is able to find the targets on it’s own if you tell it hostname of the iSCSI server. No more annoying URL typing!

The diskmanager is ready to create partitions on the iSCSI device.

Explorer view of the formatted (NTFS) testvolume.

Speeds:

iSCSI performs very fast – in my tests on real computers (no VMs used for this test), I get around 100MByte/s by copying huge files (4.5GB ISO of Fedora) from the PC/Mac to the target. Max peak was around 200MByte/s from the RAID5 in the server to the RAID0 in the PC itself, but after a few seconds the speed comes down to the already mentioned peak of around about 100MByte/s.

This speed is VERY good and a bit faster than my Samba2 experiences or on the same level.

Usage:

The normal user would ask what is the benefit of using a SAN device instead of building in a normal HDD into the computer or using a NAS or use a Windows Home Server.

SAN is made for enterprise class networks like huge companys. I had explained this to a friend last Friday and I think it was a good example:

A company with 100 users uses a program that needs direct harddisk access. Instead of buying 100x 80GB HDDs (for 40$ each), they build a huge SAN server with some TByte of storage. No user would ever use the full 80GB of the storage, and so they need to buy less storage capacity overall. Additional benefits are the usage of a secure RAID5 or RAID6 (for example) storage. It’s A LOT easier to backup, because none of the 100 user PCs or ThinClients has to run over night for a backup.

I think those points are very good reasons for iSCSI usage. Additional, the management is way easier than using a shared SMB/NFS or AFP storage, because you don’t need to fight with rights.

If I would do video editing and need some very large devices (and I speak of 20,30,40TByte), I would go for iSCSI on a reliable server.

Also, Mac users would like to use iSCSI for Time Machine.

There are a lot features inside the Enterprise Server that I haven’t tested. Ok – I’ve used the RAM drive, but it just makes sense for temp. databases and scratch stuff in my mind. Maybe a 8GB scratch over the network is not bad for Photoshop? Well, sorry – but that would be mostly slower than your local harddisk.

Prices:

While there is a limited (2TB max.) version of StarWind Server for free, the regular server costs 600$ while the Enterprise server costs 2995$. A lot money, but if you want to use it in an Enterprise network, I think it’s really worth it. The absolute pro is, that you would use it in a heterogeneous network instead of adding a SUN based Solaris/OpenSolaris distribution to use iSCSI. You can easily backup the virtual drives, because – if you use file images – they are just that: simply files. Oh! Before I forget: You can also create Snapshot volumes. How cool is that?

Useful link: iSCSI on Wikipedia (EN)

Tweet This Post Digg This Facebook Reddit

Hi guys,

since Leopard, Apple has decided to add the soft shadows around to the screenshots, when you do a window screenshot by pressing Shift+CMD+4 and then the Spacebar.

Now, since Snow Leopard (I think it just started here, because I never recognized this earlier) screenshots will be cut on the right hand side if you do a screenshot of a window that is around about 90% or more of the screen width.

Let me show you an example of doing a screenshot on Safari in Snow Leopard.

The fullscreen screenshot (done by pressing Shift+CMD+3) looks like this:

Now, when I take a screenshot of the Safari window, it results in this:

As mentioned before, not sure if this is new, but I find this very annoying. If a reader using Leopard still may test it on his/her own and leave a comment, that would be nice.

Ciao
Dennis

Tweet This Post Digg This Facebook Reddit

Hi guys,

since sunday I’m working on a new project in the late hours of the days. Since I moved to the new home office, I’ve no own ISDN connection in this room which means: I’ve no telephone available in the regular way. Sure, I have this cordless ISDN phone which I take to the home office every day at the moment, but it could be so much more comfortable. I mean – hey – I’ve an Cisco 7960 IP Phone sitting on my desk and I’ve a brand new GN Netcom “Jabra” USB Headset which I mainly bought a few days ago, when my new Logitech Headset broke down, for my daily work from the home office.

So, the plan was to create an Asterisk server inside one of the machines in the rack to put my to the ISDN line using the SIP phones (Cisco 7960) as well as the softphones on the different computers.

First, I thought I just need my few weeks back bought HFC ISDN card for everything and that I could use it on my DELL PowerEdge 1650. Well, that doesn’t work, because it’s a 3.5V card and the PCI-X slots on the DELLs riser card are made for 5V cards. Too bad. My next idea was to use it in the main fileserver which currently runs Windows Server 2008 x64 Standard Edition with some nice VMs. Of course there was NO driver available for Windows Vista/2008 64bit and I can’t link this card to a VM. Hm.. so the only way to go on with my project was to create another server for the rack.

So I grabbed my AMD64EE and put it inside my “beige” 19″ 4U case which was not used before. Hint for all of you who want to buy a new case for 19″: Pay more! Using those <100 Euro cases it HELL! My next case would be a SuperMicro or Chenbro, even if they are a "bit" more expensive. Anyway.

I've put on Debian Lenny as system OS and installed Asterisk following a bunch of tutorials I found on the net. The Asterisk itself was fast set up, but the configuration is so text-based, that I've installed Asterisk-GUI which works nice now

After spending some hours over 2 evenings & nights, my internal Asterisk works now. Most annoying thing was to make the Cisco IP Phone work with Asterisk, but in the end I solved it. Currently, I can make calls from any PC/Mac to this phone and vice versa and of course from any PC to PC or Mac to Mac or... you get the idea

But that's not what I initially wanted. I want to pick up the phone and call someone from outside. Well, that's a bit complicated with Asterisk, especially if you want to use your old ISDN phones with Asterisk.

What you need and how it should look alike is shown in the Visio drawing I've done yesterday to get an overview (click to show it in full size):

The current state is, that the connection between the ISDN and the Asterisk does not yet work. I’ve already put in both ISDN cards (1x HFC / 1x AVM Fritz), but the HFC is currently not used. First challenge is to make the ISDN connection work to the outsite. The rest is luxus

Here’s a link to an older video or playing around the first time with Asterisk and the phone, but it mainly looks like the current setup and – yeah – I think some of you know this ringtone

Link to the QuickTime Video

Have a nice day!

Ciao
Dennis

Useful links:
- Cisco 7960x on Asterisk
- Sources for your 7960x like ringtones
- Das Asterisk Buch (German)

Tweet This Post Digg This Facebook Reddit